Trans. Assoc. Comput. Linguistics 2021

89 papers
In 2021, Transactions of the Association for Computational Linguistics (TACL) published 112 papers, with a strong emphasis on the reliability and evaluation of text summarization. Research themes focused on factual faithfulness, the trade-off between faithfulness and abstractiveness, and methods for summarizing without human annotations. Other prominent areas of investigation included balanced data sampling for multi-task learning and the dynamics of multimodal fusion mechanisms. The year’s output also highlighted advancements in multilingualism and model robustness. Key contributions included the MKQA benchmark for multilingual open-domain question answering and analyses of how morphology affects language modeling. Technical developments featured model compression via causal effect estimation, decontextualization techniques to make sentences stand-alone, and edit-based transformers for machine translation. Prolific authors contributing to these developments included Graham Neubig, Yoav Goldberg, and Mirella Lapata. Collectively, the research reflected an interest in improving systematic generalization and the semantic compatibility of language models across diverse linguistic contexts.

Top Research Directions

Research directions most represented by Trans. Assoc. Comput. Linguistics 2021 papers.

The Language Bias of English Centric Benchmarks 3
Mitigating Data Scarcity in Low Resource Language Processing 3
Factual Consistency in Abstractive Summarization 2
Overcoming the Data Annotation Bottleneck in Language Processing 2
Why Standard Training Budgets Underestimate Model Potential 1
Punctuation as a Structural Signal for Natural Language Processing 1
Standardized Protocols for Controlled Neural Architecture Evaluation 1
Decoupling Knowledge Storage from Model Parameters 1
Correlating Predictive Uncertainty with Actual Model Error 1
The Failure of Structural Invariance in Language Models 1
Statistical Association Measures for Lexical and Semantic Discovery 1
Bridging the Local to Global Interpretability Gap 1
Information Theoretic Alignment for Multimodal Image Registration 1
Mitigating the Complexity of Spoken Dialogue System Development 1
Predictive Inference via Bayesian Model Averaging 1

Top Authors

AuthorPapers
Graham Neubig5
Yoav Goldberg5
Dan Roth3
Michael Collins3
Mirella Lapata3
Ryan Cotterell3
Yanai Elazar3
Antonios Anastasopoulos2
Benjamin Van Durme2
Daniel Deutsch2
Daniel Khashabi2
David Grangier2
Elias Stengel-Eskin2
Eunsol Choi2
Hinrich Schuetze2
Jennimaria Palomaki2
Jonathan Berant2
Matthew Lamm2
Noah A. Smith2
Pengtao Xie2

Impactful Papers

Papers from Trans. Assoc. Comput. Linguistics 2021 that shaped research clusters.

Influential

Papers

Title
Multimodal Pretraining Unmasked: A Meta-Analysis and a Unified Framework of Vision-and-Language BERTs
Narrative Question Answering with Cutting-Edge Open-Domain QA Techniques: A Comprehensive Study
Neural Event Semantics for Grounded Language Understanding
Neural Modeling for Named Entities and Morphology (NEMO2)
Neural OCR Post-Hoc Correction of Historical Corpora
On the Difficulty of Translating Free-Order Case-Marking Languages
On the Relationships Between the Grammatical Genders of Inanimate Nouns and Their Co-Occurring Adjectives and Verbs
PAQ: 65 Million Probably-Asked Questions and What You Can Do With Them
Parameter Space Factorization for Zero-Shot Learning across Tasks and Languages
ParsiNLU: A Suite of Language Understanding Challenges for Persian
Partially Supervised Named Entity Recognition via the Expected Entity Ratio Loss
Planning with Learned Entity Prompts for Abstractive Summarization
Pretraining the Noisy Channel Model for Task-Oriented Dialogue
Provable Limitations of Acquiring Meaning from Ungrounded Form: What Will Future Language Models Understand?
QED: A Framework and Dataset for Explanations in Question Answering
Quality at a Glance: An Audit of Web-Crawled Multilingual Datasets
Recursive Non-Autoregressive Graph-to-Graph Transformer for Dependency Parsing with Iterative Refinement
Reducing Confusion in Active Learning for Part-Of-Speech Tagging
Relevance-guided Supervision for OpenQA with ColBERT
Revisiting Few-shot Relation Classification: Evaluation Data and Classification Schemes
Revisiting Negation in Neural Machine Translation
Samanantar: The Largest Publicly Available Parallel Corpora Collection for 11 Indic Languages
Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLP
Self-supervised Regularization for Text Classification
Sensitivity as a Complexity Measure for Sequence Classification Tasks
SOLOIST: Building Task Bots at Scale with Transfer Learning and Machine Teaching
Sparse, Dense, and Attentional Representations for Text Retrieval
Structured Self-Supervised Pretraining for Commonsense Knowledge Graph Completion
Sub-Character Tokenization for Chinese Pretrained Language Models
SummEval: Re-evaluating Summarization Evaluation
Supertagging the Long Tail with Tree-Structured Decoding of Complex Categories
There Once Was a Really Bad Poet, It Was Automated but You Didn't Know It
Towards Question-Answering as an Automatic Metric for Evaluating the Content Quality of a Summary
Unsupervised Abstractive Opinion Summarization by Generating Sentences with Tree-Structured Topic Guidance
Unsupervised Learning of KB Queries in Task-Oriented Dialogs
VILA: Improving Structured Content Extraction from Scientific PDFs Using Visual Layout Groups
Weisfeiler-Leman in the Bamboo: Novel AMR Graph Metrics and a Benchmark for AMR Graph Similarity
What Helps Transformers Recognize Conversational Structure? Importance of Context, Punctuation, and Labels in Dialog Act Recognition
WikiAsp: A Dataset for Multi-domain Aspect-based Summarization

« prev Page 2/2